Language nudges Art

« previous post | next post »

Following up on Wednesday's Dinosaur comics post, Lera Boroditsky wrote:

Mark – Thanks for suggesting an important new research direction for my lab in your post yesterday.  Grammatical sexual orientation markers are definitely the next frontier.

In exchange, I offer an essay by Jakobson, which is brilliant, and not just because it begins with a meditation on the meaning of cheese. We recently tested his claim that personification in art correlates with grammatical gender.

The essay in question is Roman Jakobson, "Linguistic Aspects of Translation", in R. A. Brower, Ed., On Translation, 1959.  Jakobson's ruminations on cheese start with a reference to Bertrand Russell:

His opinions about grammatical gender and personification are here:

And the recent test that Lera mentions was published as Edward Segel and Lera Boroditsky, "Grammar in Art", Frontiers in Cultural Psychology, 2011. Its abstract:

Jakobson (1959) reports: “The Russian painter Repin was baffled as to why Sin had been depicted as a woman by German artists: he did not realize that ‘sin’ is feminine in German (die Sünde), but masculine in Russian (rpex).” Does the grammatical gender of nouns in an artist’s native language indeed predict the gender of personifications in art? In this paper we analyzed works in the ARTstor database (a digital art library containing over a million images) to measure this correspondence. This analysis provides a measure of artists’ real-world behavior. Our results show a clear correspondence between grammatical gender in language and personified gender in art. Grammatical gender predicted personified gender in 78% of the cases, significantly more often than if the two factors were independent. This analysis offers a new window on an age-old question about the relationship between linguistic structure and patterns in culture and cognition.



40 Comments

  1. Ryan North said,

    March 11, 2011 @ 8:26 am

    Fascinating! I actually just posted today's comic, which is based almost entirely on Lera Boroditsky's Sex, Syntax and Semantics paper. Clearly the next step is that all papers be published in the form of talking dinosaur comics.

  2. The Ridger said,

    March 11, 2011 @ 9:06 am

    In a translation class I posted a bit from the otherwise brilliant Russian translation of Terry Pratchett's Hogfather, in which Death (playing the part of the Hogfather (the Discworld's Santa Claus analog) prompts a child at a department store with "what do you say?" In the Russian, this is rendered as "Что-что?" не понал Смерть or "What did you say?" Death asked in confusion. (The child's words were childishly mispronounced in both versions.)

    My point was that the translator had missed the conventional meaning of "what do you say?" (to elicit a 'please') and instead rendered it literally. But what my students focused on was that Death (Смерть) was masculine in the Russian version when clearly смерть is a feminine noun and thus Death, too, should be a woman. (Utterly impossible for the translator, as Death is specifically male in Pratchett's books.)

  3. Leonardo Boiko said,

    March 11, 2011 @ 9:44 am

    I normally don’t notice the genders in my native Portuguese; they feel much milder than, say, calling a country or a ship “her” in English. However, foreign anthropomorphisms force the genders to come to consciousness. I feel nothing weird about Gaiman’s (female) Death, but Pratchett’s male Death keeps inflicting strange sentences to the poor translator.

  4. Tadeusz said,

    March 11, 2011 @ 9:56 am

    The arbitrariness of the linguistic sign is not taken into account when discussing such gender/sex mismatches in various cultures/languages, and translators do have quite a lot of devices to use when in need. For example, Смерть (Smert') can be treated as a fancy proper name (cf. the name Maria as a man's name), or the translators can produce, in Russian and other languages with a well-developed morphological system, a derivative ending with a suffix that would change the gender to masculine, (various augmentatives), etc.
    For many theoreticians the only acceptable type of translation seems to be literal translation (whatever that is, most often it is not defined at all), which is not what you can find in translated texts.

  5. John Cowan said,

    March 11, 2011 @ 11:24 am

    Here's Quine on the occasional advantage of grammatical gender, from Quiddities s.v. "gender":

    [...] In the original meaning a gender is simply a kind, irrespective of sex. The word comes from the ablative genere of the Latin genus via the French genre. Nouns and adjectives in our familiar languages divide into kinds, or genders. We do indeed associate genders with sexes by calling them masculine and feminine, and for a good reason: nouns for males are mostly masculine, and nouns for females. At the time of the original Indo-European language, our ancestors were apparently much preoccupied with sex differences, and projected them, in an animistic spirit, all across nature.

    Such, perhaps, was the irrational origin of gender as we know it. But its survival value is an independent matter, and resides in the utility of gender in sorting out anaphora, or cross-references. In complex sentences there is a continual threat of ambiguity as to which of two nouns is meant to be the antecedent of a given pronoun. We are told that

    He removed the manuscript from the briefcase and cast it into the sea

    and are left wondering whether he cast the manuscript or the briefcase. In French, all is clear:

    Il retira le manuscrit de la serviette et le/la jeta dans le mer

    Manuscript, le; briefcase, la. As luck would have it, the two nouns differ in gender. And luck does have it thus, often enough to matter.

  6. Püppi said,

    March 11, 2011 @ 11:26 am

    When I had to translate a Catalan Death to German, I just called her "Frau Tod". Not a problem.

    I hope "rpex" is an OCR effect, or do linguists transliterate by shape now?

  7. Freddy Hill said,

    March 11, 2011 @ 11:28 am

    Yes, of course for this Spanish guy death is a woman. It actually makes thinking about the whole business less distasteful.

    For us, cars are guys, which probably explains why we don't make Maseratis or Lamborghinis and the Italians, who think of cars in the feminine (la macchina!) make the sexiest cars in the world.

    I wonder, though. Do English speakers think about things in terms of gender even if there is no gender in English? Is the moon a girl, or a knife a boy?

  8. Dan said,

    March 11, 2011 @ 11:55 am

    @Freddy: anthropomorphisms and talking animals in English pick up the same background assumptions you would make about non-specifically-identified humans ("the average New Yorker", etc). So, in America, they're generally male (and white, etc) by default, unless they have some explicit indication otherwise.

  9. Colin Reid said,

    March 11, 2011 @ 12:36 pm

    Where this gets really weird for me is animals, as they clearly have a biological gender, but the way this manifests in the gender system of a language can be a bit complicated. Yes, many languages have different nouns for the male and the female of the most useful domestic animals, but often one of these is preferred in a generic context, and for most wild animals there's only one word. For instance, when talking about cats, French and German can both distinguish between a tom and a queen (chat/chatte, Kater/Katze). But in French, the male form is also used as a generic word, like 'cat' in English, while in German, the default is the female form. The word for 'mouse' in both languages is feminine and as far as I know there is no single word that means 'male mouse' in either language. Does this mean fictional mice tend to be female in both languages, while fictional cats tend to be male in France and female in Germany? How does this affect people's perception of Tom and Jerry cartoons? (Jerry in particular – the character is canonically male, but there is nothing to suggest this in most of the cartoons.)

    @Freddy Hill: In German, when cars are referred to by make, they are masculine, based on the word 'Wagen' (='vehicle') (e.g. 'ein schwarzer Mercedes', even though Mercedes is a girl's name), but nowadays the more common word for 'car' generically is the neuter 'Auto'. I think English speakers tend to make abstractions and inanimate objects female by default if they want to personify them (though most things aren't typically personified), whereas animals and generic people are often assumed male by default. Typical formal usage at the time of WWII would be to refer to 'the enemy' as 'he' but 'Germany' as 'she'.

  10. phosphorious said,

    March 11, 2011 @ 1:06 pm

    Is there something vaguely whorifian going on here? If your language forces you to ascribe genders to things, then you see objects as being either male or female?

    [(myl) Welcome to the conversation. You could start with the collected works of Lera Boroditsky, or the Economist's recent debate on the subject. Or are you only ironically pretending to be just waking up in the middle of things?]

  11. Lugubert said,

    March 11, 2011 @ 1:23 pm

    Years ago, in beginners' Religious Studies, I challenged a textbook that stated, "Philo may have preferred Logos because it is masculine in Greek, whereas Wisdom [Sophia] is feminine."

    I asked a language website, "The beginning of the Gospel according to St. John tells us that “In the beginning was the Word” (KJV). The word Word is logos in the Greek text, and it has a much wider meaning than the linguistic one: reason, mind, wisdom etc. I have repeatedly come across a theory which states that the word sophia (wisdom), which is closer to the consensus meaning of the verse, was rejected, because it has feminine gender, and logos, being masculine, had to be used, as it points forward to Jesus."

    One comment: "Anyway, in Spanish, we have no problem to address him as "La Palabra" (that is, feminine)."

    Another: "I see no reason at all why a feminine word such as wijsheid (wisdom) or a neuter word such as woord (word) could not both be applied to a male person. So the theory seems very unlikely to me too.

    But Ellegård: "Jesus – One Hundred Years Before Christ", p.84 n.24

    "The Hebrew term for Wisdom, hokmah, is grammatically masculine, while Greek sophia is feminine. Hence some writers preferred the Greek masculine words, nous (reason) and logos (word, order)."

    Summarized, I think that most people can and will make a distinction between grammatical gender and natural gender.

  12. Lugubert said,

    March 11, 2011 @ 1:28 pm

    I should perhaps have added that Alvar Ellegård was a university professor of English for many years, and contributed substantially to the start of a linguistics department in Göteborg, Sweden.

  13. Jean-Sébastien Girard said,

    March 11, 2011 @ 1:35 pm

    @Ridger and Leonardo Boiko, Pratchett's French translator actually got out of it with what I thought a very clever approach: he turned it into a running gag. Every first appearance of Death in a book will include a footnote to the effect of his maleness. They are actually perfectly in the tone of a typical Pratchett footnote, too. By the latter books, they go something like "We are not going to bother explaining anymore".

  14. Keith said,

    March 11, 2011 @ 1:40 pm

    @Colin Reid,
    "In German, when cars are referred to by make, they are masculine, based on the word 'Wagen' (='vehicle') (e.g. 'ein schwarzer Mercedes', even though Mercedes is a girl's name), but nowadays the more common word for 'car' generically is the neuter 'Auto'."

    A similar situation in French, where makes are masculine ("mon Citroën", "mon De Dion-Bouton"), whereas the general terms "voiture" and "auto" are now both feminine. However, I remember reading in an old Larousse dictionary (probably from around 1912) that the gender of "auto" was at that time not fixed and could be either masculine or feminine.

    K.

  15. Jerry Friedman said,

    March 11, 2011 @ 1:53 pm

    @Freddy Hill: But is death la comadre Sebastiana in Spain, as she is here in New Mexico?

    @Lugubert: I'm pretty sure chokhmah is feminine in Hebrew, contrary to your quotation from Ellegård. One of the few things I more or less remember is the praise of wisdom from Proverbs: `etz chayim hi' l-machaziqim bah, "a tree of life is she to them who hold strongly to her". And I checked it in modern Hebrew too.

  16. James C. said,

    March 11, 2011 @ 2:34 pm

    If Roman Jakobson didn’t think of it then Edward Sapir did. And vice versa.

  17. Coby Lubliner said,

    March 11, 2011 @ 2:52 pm

    I am still curious about the semantic significance of Lera Boroditsky's use of the masculine form of her surname. Does she want Russians (and other Slavs) to think of her as male? Or are surname gender markers not a part of "the next frontier"?

    And what's this about "sexual orientation markers?" Are there words marked for homosexuality?

    [(myl) "Sexual orientation markers" is a joking reference to the comic strip reproduced here.]

  18. Ellen K. said,

    March 11, 2011 @ 2:55 pm

    @Coby Lubliner. Did you miss my comment to you earlier in the thread regarding Lera Boroditsky not using the feminine form of her name? You didn't address my suggestion as to a possible, and quite plausable, reason for using the form she does.

  19. Jerry Friedman said,

    March 11, 2011 @ 2:56 pm

    @Coby Lubliner: In my limited experience, most Russian-born women in the U.S. use the masculine form of their surname.

    "Sexual orientation markers" was a joke based on the previous Dinosaur Comic, which MYL posted here.

  20. GeorgeW said,

    March 11, 2011 @ 3:20 pm

    Lugubert & Jerry: The Hebrew word for wisdom חָכְמָה is feminine.

  21. GeorgeW said,

    March 11, 2011 @ 3:37 pm

    In Arabic a single member of a collective is grammatically feminine. When my native Arabic speaking wife sees an animal or bird, she will refer to it as 'she' in English. My English mind immediately asks, how do you know?

    If I know the sex of an animal I use 'he' or 'she.' if I don't, like a bird in a tree, I use it.

    Bottom line: My wife selects grammatical gender (from her 1st language), I select gender based on biological sex (from my 1st language).

  22. Alain Turenne said,

    March 11, 2011 @ 4:26 pm

    @Keith:
    We do say "ma Citroën", "ma De Dion-Bouton", not "mon". I think either "voiture" or "auto", both definitely feminine nowadays, is implied, hence the "ma". It's true that we say "mon auto" and "mon Alfa-Roméo" but that's because of a pronunciation rule when the name begins with a vowel. I have no doubt that the Alfa's owner thinks of it (her?) as "ma voiture".

  23. Ellen K. said,

    March 11, 2011 @ 5:30 pm

    Oops… make that my comment in the comments on the other post. I was confused by him asking the same question in the comments to two different posts.

  24. marie-lucie said,

    March 11, 2011 @ 6:14 pm

    Keith: A similar situation [to German] in French, where makes are masculine ("mon Citroën", "mon De Dion-Bouton"), whereas the general terms "voiture" and "auto" are now both feminine

    I don't know where you get your quotes – I have never heard anyone say "MON Citroën", etc – French cars are feminine (therefore "MA Citroën") even if the brands have the names of male manufacturers. My father used to have a Toyota: every Toyota of the same model (at least) had writing on the back announcing "MA Toyota est fantastique" (capitals added)(this was a literal translation from English – it would have been more natural to say "ELLE est fantastique, MA Toyota").

    Perhaps you are thinking of "MON Opel": but here the masculine-seeming "mon" is the allomorph of the feminine before a vowel (as in "MON amie" or "MON assiette", for instance), so one could say without changing gender: "ELLE est fantastique, MON Opel".

    (La) voiture means "carriage", and automobile was first an adjective, so une voiture automobile was literally a "self-moving carriage" as opposed to une voiture à chevaux" (horse-drawn carriage). When detached from its now redundant noun, automobile kept its feminine gender while becoming a noun.

  25. marie-lucie said,

    March 11, 2011 @ 6:16 pm

    (I had not seen Alain Turenne's comment, which makes the same point).

  26. Ava Berlin said,

    March 11, 2011 @ 7:08 pm

    I'm curious: Is Boroditsky agnostic about mechanism? How does she propose these gendered associations come about? I can find no concrete proposal of this in any of her work on this question.

    Similarly, in her recent paper on metaphor, she finds that metaphors "appear to instantiate frame-consistent knowledge structures and invite structurally consistent inferences." Her empirical findings are certainly notable, but as a theoretical claim, this doesn't actually distinguish between any of the proposals on metaphor that have been made in philosophy (e.g., between Black and Davidson and Moran and so on).

    For this reason, I tend to find her output frustrating – she catalogs captivating phenomena without giving any clear indication of what kind of insight they're supposed to give us. Indeed, she does't conduct experiments that would allow her to distinguish between competing theoretical claims.

    So "language shapes thought – Sometimes. On occasion." Cool; we've seen a litany of results suggesting as much. But how and why? What are those constraints and what do they tell us about language processing, more broadly?

    I can't figure out why she's holding back on us. It's almost like she doesn't see the workings of language as a proper object of study, but instead would make a career of collecting the curious and beautiful. She's like the pint-sized Indiana Jones of neo-Whorfian research.

  27. David Eddyshaw said,

    March 11, 2011 @ 7:49 pm

    @John Cowan:

    I think Quine was just plain wrong in supposing that the origin of grammatical gender is to be sought in our linguistic forbears being "much preoccupied with sex differences, and project[ing] them, in an animistic spirit, all across nature." Pretty implausible scenario a priori anyway IMHO.

    Apart from anything else, it's not difficult to find near-synonyms with non-human reference that have different grammatical genders even in the same language, and in many if not most systems the phonetic shape of words can often correlate pretty well with gender.

    Whole books have been written by actual experts on this.

  28. maidhc said,

    March 12, 2011 @ 4:31 am

    If the same word has different genders in two different Indo-European languages, is it always the case that they came from different roots, as "tod", "mort" and "smert" appear to? ("smert" looks more like "mort" than it does like "tod", I guess.)

    Are there cases where a parallel word has different genders in different languages? If so, it must be that one of them changed, which would be an interesting thing to contemplate.

    I've read an alternative theory about gender in Indo-European, other than the early IEs being sex-obsessed. That would be that the gender system represents the decay of a more elaborate noun classification system. Such classification systems are found (I believe, but I'm not a linguist, so I welcome enlightenment) in languages such as Bantu, Japanese and some American Indian languages. For example, there might be a special grammatical category for objects which are more or less round. In Bantu languages, Wikipedia reports up to 22 noun classes (http://en.wikipedia.org/wiki/Noun_class).

    Of course, speculation about what proto-IEs spoke before they spoke IE is going to be hard to justify…

    IE had a dual number, which indicates that the IEs had some interests in things that came in pairs. It has been suggested that nomadic pastoralists might encounter more paired things than we do in modern life.

    The dual number lingers in Irish and Hiberno-English, where a something that is normally part of a pair is called a half-something. For example, where one might say "I was listening to it with one ear", in Hiberno-English you would say "I had a half-ear on it", which is a literal translation of the Irish phrase.

    The Irish Consul here spoke that phrase to me once, and it warmed my heart that the Irish diplomatic service would include the promotion of regional usages among its many other duties.

    Not to be confused with a "half-door", which really is half a door.

  29. Colin Reid said,

    March 12, 2011 @ 7:50 am

    @maidhc: It's pretty common for cognate words to have different genders. Borrowed words are frequently assigned genders not based on their gender in the original language, but genders often don't match up across languages even for 'indigenous' words. In fact there are instances where a word is exactly the same in different dialects of the same language, but the gender is different. For instance 'Butter' is feminine in standard German, but I've heard it's often neuter in Austria.

    Then again there are also words distinguished *only* by gender. For instance 'la chèvre' is 'the goat' whereas 'le chèvre' is 'goat's cheese'. I imagine these don't get switched around in closely-related languages.

  30. Moacir said,

    March 12, 2011 @ 8:48 am

    @Coby Lubliner: Lithuanian isn't Russian, etc., of course, but it has 3-4 last name forms: a masculine, a feminine for unmarried women, a feminine for married women, and a (newish) feminine neutral form. I have had friends from Lithuania come to the US and not understand how it is that so many US-born Lithuanian women have masculine forms of their last names. But it's just not done in the US, as Ellen noted. So I did a test:

    In my list of Facebook friends, there are 310 women who have Lithuanian last names that are morphologically compliant (they can nicely be formed into the four forms above). Of the women who are primarily based in the Americas or Australia (273), a full 45 use the "correct" form of their last name. Of that 45, about 15 (at least) were born in Lithuania (or the Lithuanian SSR) and grew up in the US/emigrated to the US. That set of women (born in Lithuania) is also present in the people who use the masculine form of their name.

    The rest of the women, who are mostly Lithuania-born and currently live/grew up somewhere in Europe (including Lithuania) have near 100% compliance with the "correct" form. The one exception has a peculiar Lithuanian name (-a ending) which may not change.

    Note that these aren't names on birth certificates. These are the names these women have chosen for themselves on Facebook, where one can play cultural patriot or not, no matter what's on the driver's license. Also, I've met all of these women through various Lithuanian cultural events, meaning that they either speak Lithuanian or have enough interest in it to have noticed that last names change forms.

    So Boroditsky's using that form of her name is, imo, as Ellen said, completely predictable, considering the time she's spent in the US (I don't know where she was born). I'll note also that of the women in my Facebook friends list, *every* woman who has created a brand around her name in the US has used the masculine form.

    I'll give an example of what moving to the US does: one Facebook friend was born in Lithuania, grew up in LA, and had the unmarried female form of her last name as her Facebook surname. She then married a Poland-born Lithuanian who grew up in Canada, though they met and have lived their entire romantic life together in LA. After marriage, she added the masculine form of her husband's name to her last name, not the married female form, which would be the linguistically "correct" formation.

  31. [links] Link salad takes out its guitar and plays | jlake.com said,

    March 12, 2011 @ 10:10 am

    [...] Language nudges Art — Language Log on linguistic gender and cheese. Mmm. [...]

  32. William Steed said,

    March 12, 2011 @ 5:58 pm

    The next step, as I see it, is to check the correlation between English anthropomorphic personifications and the words' historical genders before English did away with most of them.

  33. John Cowan said,

    March 13, 2011 @ 1:44 am

    David Eddyshaw: I think Quine was just plain joking about that part, but you may be in fact right that he was in fact wrong. He isn't wrong about the rest, though.

  34. Bob Ladd said,

    March 13, 2011 @ 5:52 am

    @maidhc: The Romance languages provide numerous cases of cognate words with different genders. A lot of these came from Latin "3rd declension" nouns ending in -e, which don't give away their gender from their form, unlike 1st declension nouns in -a (almost invariably feminine) and 2nd declension nouns in -u/-o (almost invariably masculine (or neuter, but these neuters almost all became masculine in Romance)). So flowers (Lat. source flore(m), masculine) are masculine in Italian but have become feminine in most of the others; teeth (Lat. source dente(m), masculine) have become feminine in French but are masculine in most of the others; the sea (Lat. source mare, neuter) is masculine in Italian and (usually) in Spanish but feminine in French and Romanian; and the Latin word fronte(m), feminine, with the literal meaning 'forehead' and the transferred meaning 'front', became masculine in French in all senses, but only in the transferred senses in Spanish and Italian. There are more examples, but these give you the idea.

  35. Geoff Nunberg said,

    March 14, 2011 @ 2:00 am

    If memory serves (hah!), I once read a recollection by André Martinet that on his first visit to New York he went to the Cloisters to see the Unicorn Tapestries and was struck to hear the beast ("la licorne") referred to as "he." I can't find the text, though…

  36. Lepock said,

    March 14, 2011 @ 11:02 am

    I don't think this result shows anything about the relationship between language and cognition. It's probably just an artifact of how artists have to manipulate how audiences respond to their work.

    Here's a scenario for you. Suppose Shakespeare wrote a play in which Sin appears. Shakespeare didn't care what gender Sin was. He flipped a coin, it came up tails, and Sin became a woman.

    Now suppose it's today and I'm writing a play with a personification of Sin. Suppose I don't think Sin has a gender; in fact, I think the whole idea of Sin having a gender is nonsensical. I'm only giving Sin a gender because of the unfortunate dearth of genderless actors to read Sin's lines.

    I know that many of my readers will have read Shakespeare's play, and many of them will be comparing my treatment of Sin to Shakespeare's as they watch. Suppose I make Sin a man. There are bound to be people in my audience who see this and think "Oh, his Sin is a man and Shakespeare's was a woman! I wonder what he might mean by that?" Now, of course, I don't mean anything by it. I don't want my audience to think about Sin's gender at all. I want them to focus on (say) the profundity of what Sin is saying.

    So I make Sin a woman just like Shakespeare did, in order to minimize the probability the audience will be distracted. I do this precisely because I think of Sin as genderless.

    It's the same general technique that leads to stereotypical characters in fiction. Suppose in one scene a duchess (a very minor character) gets into her limousine, and her driver is holding the door open for her. Suppose I know plenty of aristocrats, and I know that they're very kind and gracious to their servants. I think a real duchess would thank her driver with a smile as she got in the car. However, my readers expect aristocrats to be arrogant and rude. I know this isn't true, but I need to get the duchess in the car so the story can keep moving. Making her gracious would defy the reader's expectations, call attention to a minor character, and just generally derail the plot. So I'll make her haughty to her driver to minimize distractions and keep the reader focused on the story. It says nothing about what I think of duchesses.

    Now you can imagine how strong the distracting effect of a mismatch between the gender of a character in a painting and the word used to describe it would be. Someone describing a male Sin in German is pretty likely to notice that mismatch. And then they just might think to themselves, "Might the artist mean anything by that?" And if I don't mean anything by giving Sin a gender, I'd be foolish to allow the viewer to think about that rather than what I really want to convey.

    Thus I think the effect can be completely explained by the artist's not thinking of objects having genders of their own, but trying to minimize the likelihood that the audience might wonder about that issue at all.

  37. David Marjanović said,

    March 14, 2011 @ 5:09 pm

    When I had to translate a Catalan Death to German, I just called her "Frau Tod". Not a problem.

    Might still be. I (native speaker of German) would wonder whether that's supposed to be Death or Death's wife.

    I suppose context clears it up quickly enough, though.

    Does this mean fictional mice tend to be female in both languages, while fictional cats tend to be male in France and female in Germany?

    Yes. They have the sex that fits their gender until proven otherwise.

    How does this affect people's perception of Tom and Jerry cartoons? (Jerry in particular – the character is canonically male, but there is nothing to suggest this in most of the cartoons.)

    Not at all. That's because both of them behave in stereotypically male ways all the time, up to and including falling in love with females that are overloaded with tertiary sexual characteristics. In other words, the "until proven otherwise" effect kicks in very quickly.

    Pratchett's French translator actually got out of it with what I thought a very clever approach: he turned it into a running gag. Every first appearance of Death in a book will include a footnote to the effect of his maleness. They are actually perfectly in the tone of a typical Pratchett footnote, too. By the latter books, they go something like "We are not going to bother explaining anymore".

    Awesome!

    If the same word has different genders in two different Indo-European languages, is it always the case that they came from different roots, as "tod", "mort" and "smert" appear to?

    Often, yes, but…

    It's pretty common for cognate words to have different genders. Borrowed words are frequently assigned genders not based on their gender in the original language, but genders often don't match up across languages even for 'indigenous' words. In fact there are instances where a word is exactly the same in different dialects of the same language, but the gender is different.

    Correct.

    For instance 'Butter' is feminine in standard German, but I've heard it's often neuter in Austria.

    No, masculine.

    I've read an alternative theory about gender in Indo-European, other than the early IEs being sex-obsessed. That would be that the gender system represents the decay of a more elaborate noun classification system.

    That would make sense, but it's not what the evidence seems to show. The extinct Anatolian languages, Hittite for example, had two genders, animate (for living beings) and inanimate. The branch that led to the rest of IE somehow developed a suffix to mark words for females, and this somehow (by sex obsession?) spread or was interpreted into other words or whatever… that's why masculine and neuter words look so similar in Latin while feminine ones are quite different, and why "who" behaves a lot like a masculine word in many IE languages the same way that "what" behaves as neuter.

    English has sort of regained the animacy distinction. Child/kid/baby can be "he" or "she" in English, but they're almost never "it". In German, they're all "it" all the time; if you want to mention "he" or "she" later on, you must choose "boy" or "girl" right at the beginning. German lacks an animacy distinction, English has one.

  38. Ellen K. said,

    March 16, 2011 @ 7:46 am

    @David Marjanović:. Is it really an animacy distinction in English when we can us "it" for animals?

  39. Food made of pressed curds said,

    March 21, 2011 @ 1:56 am

    [...] of fields, such that I was both fully prepared for Language Log's recent discussion of a wee spat between Bertrand Russell and Roman Jakobson, and ready to discuss, briefly, with friends the significance of invitations to dinner in Roman [...]

  40. Of female sin and the lesbian moon - G7Finance.com - Special Finance Resources said,

    April 16, 2011 @ 9:09 pm

    [...] in German, and so usually depicted as a woman in German art. The effect is strong: it holds true 78% of the time. And this all came out of the possibility of a lesbian French moon in Dinosaur Comics. Happy [...]

RSS feed for comments on this post